Download the entire proceedings as an Adobe PDF - Eastern Snow ...
Download the entire proceedings as an Adobe PDF - Eastern Snow ...
Download the entire proceedings as an Adobe PDF - Eastern Snow ...
You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
correspondence between simulated <strong>an</strong>d observed spring Q suggests <strong>the</strong> model—to some degree—is<br />
accounting for processes connecting <strong>the</strong> snowpack <strong>an</strong>d spring river flow, e.g. sublimation, rainfall,<br />
<strong>an</strong>d soil infiltration.<br />
To better underst<strong>an</strong>d <strong>the</strong> covari<strong>an</strong>ce between SWE <strong>an</strong>d Q, alternate comparisons were made<br />
using PWBM/ERA-40 monthly SWE <strong>an</strong>d <strong>an</strong> estimate of when thaw is <strong>as</strong>sumed to have occurred.<br />
The month of thaw (TM,withTM− 1, <strong>an</strong>d TM+ 1 indicating <strong>the</strong> month preceding <strong>an</strong>d postceding<br />
<strong>the</strong> thaw month, respectively) w<strong>as</strong> determined with a step edge detection scheme applied to SSM/I<br />
brightness temperatures (McDonald et al., 2004). Then, SWETM becomes monthly b<strong>as</strong>in SWE<br />
during TM (or TM− 1), <strong>an</strong>d QTM is discharge in month TM. These alternate comparisons (across<br />
all 179 b<strong>as</strong>ins) are defined (a) SWETM vs. spring Q, (b)SWETM-1 vs. spring Q, (c)SWETM-1 vs.<br />
QTM+1, (d)SWETM-1 vs. QTM+1,2. R 2 s are highest for alternate comparison (b), which compared<br />
SWE in <strong>the</strong> month before thaw (TM − 1) with spring (April–June) Q (Table 1). Yet, despite <strong>the</strong><br />
fact that <strong>the</strong> me<strong>an</strong> R 2 across western Eur<strong>as</strong>ia improves from 0.15 (using default PWBM/ERA-40)<br />
to 0.38 (alternate comparison b), little difference is noted with <strong>the</strong> remi<strong>an</strong>ing alternate comparisons<br />
<strong>an</strong>d o<strong>the</strong>r regions.<br />
L<strong>as</strong>tly, we scaled spring Q using a factor S, whereS = PWBM monthly snow melt–runoff ratio,<br />
with 0 < S < 1. Then, snowmelt Q each month is Qs =Q·S. Each occurrence of QS w<strong>as</strong> <strong>the</strong>n<br />
summed resulting in a total QS each spring, for each b<strong>as</strong>in. Using QS in place of <strong>the</strong> default Q (<strong>an</strong>d<br />
PWBM/ERA-40 SWE), we note a decre<strong>as</strong>e in agreement across e<strong>as</strong>tern Eur<strong>as</strong>ia, with no ch<strong>an</strong>ge<br />
across most of <strong>the</strong> domain. And although SWE from simulations with ERA-40, in general, explains<br />
more th<strong>an</strong> a third of <strong>the</strong> variation in Q, a large proportion of <strong>the</strong> inter<strong>an</strong>nual variability is not<br />
due to SWE variability. When considering <strong>the</strong>se results, it is interesting to note that Lammers et<br />
al. (2006) recently found that <strong>an</strong>nual simulated discharge across Al<strong>as</strong>ka (drawn from three separate<br />
models) w<strong>as</strong> in poor agreement with observed discharge data between 1980–2001. Better agreements<br />
across northwestern North America, e<strong>as</strong>tern Eur<strong>as</strong>ia (EE in Figure 3), <strong>an</strong>d parts of western Eur<strong>as</strong>ia<br />
(WE) in this study are attributable to relatively higher snowfall rates <strong>an</strong>d a greater inter<strong>an</strong>nual<br />
variability in spring discharge (Figure 4b). Conversely, <strong>the</strong> region of e<strong>as</strong>tern Eur<strong>as</strong>ia with numerous<br />
negative correlations is characterized by low spring discharge variability. Delays in snowmelt water<br />
reaching river systems, which c<strong>an</strong> be signific<strong>an</strong>t (Hinzm<strong>an</strong> <strong>an</strong>d K<strong>an</strong>e, 1991), are likely <strong>an</strong> additional<br />
influence on <strong>the</strong>se reported correlations. For large arctic b<strong>as</strong>ins, comparisons between snow storage<br />
<strong>an</strong>d discharge volume are complicated by <strong>the</strong> large temporal variation in b<strong>as</strong>in thaw <strong>an</strong>d <strong>the</strong> delays<br />
in snowmelt water reaching <strong>the</strong> gauge. More me<strong>an</strong>ingful comparisons between spatial SWE <strong>an</strong>d<br />
river discharge are possible through <strong>the</strong> use of hydrograph separation to partition discharge into<br />
overl<strong>an</strong>d <strong>an</strong>d b<strong>as</strong>eflow components. This, however, requires <strong>the</strong> use of daily discharge data which<br />
are more limited for <strong>the</strong> P<strong>an</strong>-Arctic region.<br />
CONCLUSIONS<br />
In our comparisons of inter<strong>an</strong>nual variations in pre-melt SWE <strong>an</strong>d spring Q, R 2 values are<br />
highest (me<strong>an</strong> of 0.25 to 0.28 over all b<strong>as</strong>ins) when PWBM is driven by ERA-40, NNR or WM<br />
climate data. Similar agreements are noted when SWE from <strong>the</strong> observed data <strong>an</strong>alysis scheme are<br />
used, which suggests that <strong>the</strong> hydrological model is capturing <strong>as</strong> much variability in <strong>the</strong> spring flow<br />
<strong>as</strong> does <strong>the</strong> observed SWE scheme. Average R 2 determined from <strong>the</strong> SSM/I SWE <strong>an</strong>d spring Q<br />
comparisons are generally low, <strong>an</strong>d a sizable majority (over 72%) of <strong>the</strong>se correlations are negative.<br />
The low variability <strong>an</strong>d magnitude is likely related to saturation of <strong>the</strong> SSM/I algorithm at high SWE<br />
129